Blitz-Dodgeball
Take home messages
- Speed of play can be quantified and used to inform about tactics
- Fast-paced, laissez-faire, Blitz-dodgeball can be a great adaptive tactic depending on the opponent
- Playing against opponents that are tactically superior, Blitz-Dodgeball can level the playing field
Laissez-faire play
I play a somewhat unconventional type of blitz-dodgeball: fast, with lots of exchanges and countering, as well as switching positions and covering the entire court. I thrive in a quick game since I am both physically fast and fast at counting ball possession and updating my estimates of success in each given situation (more on how to quantify metacognitive decision making in this post. For decision-making depending on ball possession, see this post). What’s more, it is a way of playing that Dodgeballers are not used to.
There are two reasons why: experience and mental aptitude. Studies show, for example, that left-handed tennis players and baseball pitchers have an advantage against opponents. This is due to the novelty of playing against them, after a lifetime of experience playing against right-handed opponents. Being forced to play a fast game when you are used to slow and methodical play can throw away years of tactical training, leaving the opponent with nothing but their pure physical prowess and technical skill. Studies also show that cognitive abilities, such as working memory, greatly affect success in fast and highly complex sports. In order to play quickly, one must be able to think quickly. Those who cannot are instead forced into making miscalculations and errors. In the end of this post, I will show just how efficient this method of playing can be.
Challenging traditions
I’ve been told that this is not the way to play dodgeball. This is not to say that everyone can and should play that way. It certainly is not for everyone, but there are some huge benefits for those who can play that way.
Before getting to the benefits, I’ll explain how I quantify playing speed. The conventional strategy in Dodgeball is to collect the balls on your side, after which the playmaker decides who throws and at which opponent(s), who have regrouped at their baseline, and then the attack occurs. This way of playing is slow and methodical, and one can expect about 1-3 balls to be thrown every 15-20 seconds. I prefer instead to throw pre-throws (attack opponents while they are attacking), make exchanges (throw at someone who is currently throwing at my teammates), do quick counter attacks, pick up loose balls that I recover across the court and then attack, etc., all while the opponents are disorganised and haven’t had a chance to regroup at their baseline. This way of improvised playing can easily result in 1-2 throws every 10 seconds.
I’ve also recorded meta-data on whether each throw is planned (decided by the playmaker and thus forced), or whether it is improvised (the player decides in the spur of the moment). This way I can compare the general playing style and which type is more successful, both for the individual player, and against which opponents. I’ve written a post (ball possession) where I show how some players are really good at improvised play but awful in terms of planned play and forced throws, and vice versa. This is one of those stats that I most frequently use, and often study hard before a competition in order to know which players I can used for planned attack, and which players to encourage to counter with me or give them free hands to improvise. Some players have really good arms, but barely get hit percentages over 15% in organised attacks. I’ve developed a method for calculating the estimated net effect of each throw; how many opponents a player is expected to eliminate versus how many teammates the opponents are expected to eliminate when they get ahold of that players ball, which will determine whether that player is expected to contribute toward a win or a loss. More on this in a future post.
To estimate the speed of playing, I count the number of events (throws in attack, and blocks, dodges and misses in defence) that occur in a given time, such as a set, and divide that value by the total time (in minutes). This way I get an estimate of how often any throw occurred. Sets where both teams largely rely on organised play will have playing speeds of around 10-12 events per minute. Those where both teams go berserk and throw as soon as they pick up a ball will be closer to 16-18.
Speed of play = Number of events / Time
- Events are throws by both teams, including invalid throws and line infringements
- Time is the time window (in minutes) during which events are counted, such as length of set or match
I’ve been told that playing that fast will increase the number of mistakes made (the patterns for the technical percentages and stable number of thrown catches speak against this), and will leave the team more exposed without any balls to defend with (the number of balls in possession do not seem to be affected by this, however there is an interaction with defence percentage). One could expect that a faster gameplay will lead to more variable results, with slow sets giving more clustered teams and fast sets giving greater spread in terms of points/min and technical percentages. This, too, is not corroborated by data. Below is a figure showing how points per minute varies with playing speed.
Sets played by the Swedish Mixed Team at the Central European Championships 2019. Green = Win, Yellow = Draw, Red = Loss.
As one can see, there is a highly significant positive correlation, with an even spread (no heteroscedasticity) across the spectrum of playing speeds. It is evident that the curve is bent, and that the improved point scoring with higher playing speeds gets gradually lower. This is expected, since faster and faster playing speeds inevitably lead to throws being taken in situations with lower and lower expected probabilities of elimination. However, at the playing speeds we had at the CEC, that peak was not yet reached, and higher speeds led to higher points/min (see hit percentage and explanation below). The outlier set against Croatia was an extremely fast set, not because of a high playing speed but because of two catches thrown and simultaneously low hit and defence percentages, leading to fast eliminations and a short set.
One must bear in mind that this pattern is specific for this team only. Other teams may well have a negative association, and playing against a team with a positive correlation they would be advised to slow down play and keep a deep frontline when attacking.
Speed of play and technical skills
A similar, but more extreme pattern of diminishing returns can be seen for defence percentage, which has an optimal play speed and quickly falls at higher speeds due to the increased vulnerability. There seems to be, for this team, a higher success rate when not all of the defensive actions are from organised play. However, after a certain speed, this type of play leaves the players more vulnerable, which diminishes the defence percentage.
Even though the defence percentage drops with higher speeds, this effect is offset by the effect on hit percentage. Hit percentage instead increases across speeds, which is a bit unexpected. This is probably due to the fact that higher playing speeds are not achieved merely by an attempt to throw more frequently, but rather those are sets with more counter attacks and exchanges, which inherently have a higher probability of elimination. I expect that the actual curve will be S-shaped, with diminishing returns at higher, forced increases, in playing speed.
Superimposing these two curves will allow the team to identify the optimal speed of play, resulting in the highest technical team skill (for Sweden, at this tournament, it is somewhere between 13,5 and 15,3 events/min.
Blitz-Dodgeball as a tactic
To show how powerful a fast playing speed is for the Swedish Mixed team, consider the results in the final against arguably the best team at the CEC, Austria. Despite having much better team composure, tactical prowess, and more experience, when the game becomes faster and organised play disappears, they stop relying on those advantages. Many Dodgeballers come from a background in handball, and I’m sure they can appreciate how confusing and disruptive it is to build up an attack against a team that plays 5-1 in defence, rather than the regular 6-0. With higher game speeds, we demolished Austria, and 60% of our point scoring is determined by playing speed. Rather than controlling the game and playing “good old dodgeball”, it was the Laissez-faire, blitz-type dodgeball that was behind much of the Swedish success at the CEC. Plainly said, Sweden is a dominating force during improvised play, and just above average when playing planned plays.
The correlations for the other teams was lower or nonexistent because our wins against them were dependent on other aspects than playing speed. Ironically, perhaps, aspects such as better organised and conventional play…
Dodgeball is a nonlinear sport, and one must know when to be adaptable and how. Playing Dodgeball against Goliath-Austria would clearly not work. The only hope of winning was to change the way the game is played and introduce Blitz-Dodgeball.
Darko unhinged
To illustrate just how much I adhere to blitz-dodgeball, consider the figure below which shows the proportion of time I spent on court in the different sets. It is clear that whenever I am on court, the speed of play is high, and when I am benched or get eliminated early, the speed of play is low. Remember that this is speed of play on the whole court, for both teams, not just my own playing speed or that of my team. It is not only that I have an anarchic playing style in attack, with lots of throws, but I also put myself in defensive situations involving calculated risks in order to fish out unsuccessful throws from the opponents which either gives us additional balls or allows my teammates to counter attack or make exchanges. This way I both increase our play speed, and force my opponents to increase their play speed.
I have developed three methods to quantify the effect that individual players have and what would happen to the game outcome if they were substituted. I explain one of them in this post, and show how much my presence influences speed of play, and how I use that information to calculate what the expected point scoring would be in my absence.
